Raafi-Karim Alidina, consultant at Included, examines why diversity is crucial when it comes to product development in the pharmaceutical and life sciences industry.
Many organizations realize the need for change. This was catalyzed by the Black Lives Matter movement and the inequalities highlighted by the pandemic. The need to be actively anti-racist has become more evident and urgent, with organizations rightly under pressure from internal and external forces. This leaves it up to leaders to tackle prejudices in processes, structures and behaviors. Many have turned to technology, with the logic that it would eliminate unconscious biases in people and create a more egalitarian and inclusive approach. This thinking, however, is wrong because these machines and algorithms are always created by people. This leads to the creation of biased technology.
Bias in technology design
Recently I went to a barbecue restaurant with my partner. When I went to the bathroom to wash my hands, everything was contactless: there were automatic taps, hand dryers, soap dispensers… even an automatic paper towel dispenser. However, as soon as I tried to use one of them, they didn’t work! No matter what I did, I just couldn’t get the sensor to detect my hands and dispense water or soap to me. Then someone else walked into the bathroom and tried washing their hands and it worked perfectly. What was the difference? Her skin was white, while mine was brown.
If you’re a person of color, you probably know that most faucets, hand dryers, and automatic soap dispensers in public restrooms don’t work as well for you as they do for your white friends.
It stems from how these products were designed and by whom they were designed. They were built on a training set that was not diverse, using images of predominantly white skinned hands, thereby incorporating this bias into how the product works. When it hits automatic faucets, hand dryers and soap dispensers, it’s an annoying problem for people of color. But the real problem arises when the same object detection system is deployed in technologies such as self-driving cars. A study at Georgia Tech University in the United States found that many autonomous vehicles in development that used these systems were significantly less good at detecting – and therefore stopping for – dark-skinned pedestrians. This technology could pose a very real threat to blacks and browns and lead to injuries and deaths. This is not the only example of coded bias and it affects all kinds of products. Using crash test dummies modeled exclusively on the average male body, Amazon’s biased hiring algorithm, To Google’s failed launch from their search for inverted images (which identified black people as “gorillas”), or even many speech recognition software that cannot accurately detect black voices (nor many accents).
Inclusion must be present in the design process, not only to avoid inconvenience, but also to avoid serious consequences for the public.
Two main problems are at the origin of this coded bias. First, the lack of diversity in product manufacturing teams. Undiversified teams are less likely to spot issues that may impact a different demographic than the one they are a part of. Second, there is a lack of inclusion in these teams. Where there are diverse teams, those from marginalized groups are often not able to speak out and be heard. This may be due to insufficient psychological security arising from the culture of the organization.
How does this impact pharma?
The pharmacy is not immune to this coded bias. For example, women have been excluded from drug trials for decades without any evidence-based justification for not including a demographic group that represents 51% of the world’s population. This has very real implications, as the doses were decided by male-dominated clinical trials, leading to women experiencing worse side effects than their male counterparts. For example, a recent UC Berkeley and UCicago study conducted in 2020 found that of 86 drugs reviewed (all approved by the U.S. Federal Drug Administration), women suffered more serious side effects. in more than 90% of cases. These side effects weren’t just headaches and nausea (although these were some of them), but major issues like seizures, hallucinations, and heart abnormalities.
The study notes that even extremely common drugs like Ambien sleeping pills have different effects on men and women – they “stay in the bloodstream of women longer than men, causing drowsiness the next morning, significant cognitive impairment. and an increase in road accidents “. It wasn’t until 2013 that people noticed and dosage recommendations declined for women.
As with coded bias in technology, the lack of diversity and inclusiveness of the team is the driving issue here. Teams made up of people with similar backgrounds are more likely to have a blind spot and less likely to notice that a clinical trial group is not representative of the population. Diverse and representative teams should always ensure that a psychologically safe environment is created so that team members from non-dominant groups can speak out and be heard if they notice a bias, blind spot, or a mistake.
What is the solution?
There must be a conscious and deliberate decision to integrate inclusion and diversity into the process used when designing new products. This includes drugs, medical devices and all methods of treatment. By taking this approach, pharmaceutical companies can minimize the risk of default biases being reflected in the final product created.
Some ways this can be achieved are by adding a:
- Inclusion review. Incorporating this into the product development process, much like a legal review, acts as a check to make sure all blind spots are covered.
- Inclusion checks at different points in the process. This takes advantage of the insight of a diverse team at specified checkpoints throughout the product design and development process. Teams can actively engage their members, asking everyone to comment specifically on any potential biases they can identify in the way the research is conducted.
- Start and end the product lifecycle by reflecting on your identities and biases. This provides an opportunity to challenge and engage with one’s own biases and those of the team as a whole. This means that the product team can start the project with an inclusive mindset before starting development. By including this at the end of the project, a retrospective learning opportunity is present to enable a culture of continuous improvement to drive future design.
These steps can help the pharmaceutical industry and all industries create more inclusive products. Every solution is active, because passive solutions will not help create the better world we seek to build. To stop making preventable mistakes that risk harming people, each of us must take active steps to embed an inclusive approach into the way we do our daily work.